GitLab-managed Terraform state (FREE)

Introduced in GitLab 13.0.

Terraform remote backends enable you to store the state file in a remote, shared store. GitLab uses the Terraform HTTP backend to securely store the state files in local storage (the default) or the remote store of your choice.

WARNING: Using local storage (the default) on clustered deployments of GitLab will result in a split state across nodes, making subsequent executions of Terraform inconsistent. You are highly advised to use a remote storage resource in that case.

The GitLab-managed Terraform state backend can store your Terraform state easily and securely, and spares you from setting up additional remote resources like Amazon S3 or Google Cloud Storage. Its features include:

  • Versioning of Terraform state files.
  • Supporting encryption of the state file both in transit and at rest.
  • Locking and unlocking state.
  • Remote Terraform plan and apply execution.

A GitLab administrator must set up the Terraform state storage configuration before using this feature.

Permissions for using Terraform

In GitLab version 13.1, at least the Maintainer role was required to use a GitLab managed Terraform state backend.

In GitLab versions 13.2 and later, at least:

  • The Maintainer role is required to lock, unlock, and write to the state (using terraform apply).
  • The Developer role is required to read the state (using terraform plan -lock=false).

Set up GitLab-managed Terraform state

To get started with a GitLab-managed Terraform state, there are two different options:

Terraform States can be found by navigating to a Project's {cloud-gear} Infrastructure > Terraform page.

Get started using local development

If you plan to only run terraform plan and terraform apply commands from your local machine, this is a simple way to get started:

  1. Create your project on your GitLab instance.

  2. Navigate to Settings > General and note your Project name and Project ID.

  3. Define the Terraform backend in your Terraform project to be:

    terraform {
      backend "http" {
      }
    }
  4. Create a Personal Access Token with the api scope.

  5. On your local machine, run terraform init, passing in the following options, replacing <YOUR-STATE-NAME>, <YOUR-PROJECT-ID>, <YOUR-USERNAME> and <YOUR-ACCESS-TOKEN> with the relevant values. This command initializes your Terraform state, and stores that state in your GitLab project. This example uses gitlab.com:

    terraform init \
        -backend-config="address=https://gitlab.com/api/v4/projects/<YOUR-PROJECT-ID>/terraform/state/<YOUR-STATE-NAME>" \
        -backend-config="lock_address=https://gitlab.com/api/v4/projects/<YOUR-PROJECT-ID>/terraform/state/<YOUR-STATE-NAME>/lock" \
        -backend-config="unlock_address=https://gitlab.com/api/v4/projects/<YOUR-PROJECT-ID>/terraform/state/<YOUR-STATE-NAME>/lock" \
        -backend-config="username=<YOUR-USERNAME>" \
        -backend-config="password=<YOUR-ACCESS-TOKEN>" \
        -backend-config="lock_method=POST" \
        -backend-config="unlock_method=DELETE" \
        -backend-config="retry_wait_min=5"

    WARNING: The name of your state can contain only uppercase and lowercase letters, decimal digits, hyphens, and underscores.

If you already have a GitLab-managed Terraform state, you can use the terraform init command with the pre-populated parameters values:

  1. On the top bar, select Menu > Projects and find your project.
  2. On the left sidebar, select Infrastructure > Terraform.
  3. Next to the environment you want to use, select the Actions menu {ellipsis_v} and select Copy Terraform init command.

You can now run terraform plan and terraform apply as you normally would.

Get started using GitLab CI

If you don't want to start with local development, you can also use GitLab CI to run your terraform plan and terraform apply commands.

Next, configure the backend.

Configure the backend

After executing the terraform init command, you must configure the Terraform backend and the CI YAML file:

  1. In your Terraform project, define the HTTP backend by adding the following code block in a .tf file (such as backend.tf) to define the remote backend:

    terraform {
      backend "http" {
      }
    }
  2. In the root directory of your project repository, configure a .gitlab-ci.yml file. This example uses a pre-built image which includes a gitlab-terraform helper. For supported Terraform versions, see the GitLab Terraform Images project.

    image: registry.gitlab.com/gitlab-org/terraform-images/stable:latest
  3. In the .gitlab-ci.yml file, define some CI/CD variables to ease development. In this example, TF_ROOT is the directory where the Terraform commands must be executed, TF_ADDRESS is the URL to the state on the GitLab instance where this pipeline runs, and the final path segment in TF_ADDRESS is the name of the Terraform state. Projects may have multiple states, and this name is arbitrary, so in this example we set it to example-production which corresponds with the directory we're using as our TF_ROOT, and we ensure that the .terraform directory is cached between jobs in the pipeline using a cache key based on the state name (example-production):

    variables:
      TF_ROOT: ${CI_PROJECT_DIR}/environments/example/production
      TF_ADDRESS: ${CI_API_V4_URL}/projects/${CI_PROJECT_ID}/terraform/state/example-production
    
    cache:
      key: example-production
      paths:
        - ${TF_ROOT}/.terraform
  4. In a before_script, change to your TF_ROOT:

    before_script:
      - cd ${TF_ROOT}
    
    stages:
      - prepare
      - validate
      - build
      - deploy
    
    init:
      stage: prepare
      script:
        - gitlab-terraform init
    
    validate:
      stage: validate
      script:
        - gitlab-terraform validate
    
    plan:
      stage: build
      script:
        - gitlab-terraform plan
        - gitlab-terraform plan-json
      artifacts:
        name: plan
        paths:
          - ${TF_ROOT}/plan.cache
        reports:
          terraform: ${TF_ROOT}/plan.json
    
    apply:
      stage: deploy
      environment:
        name: production
      script:
        - gitlab-terraform apply
      dependencies:
        - plan
      when: manual
      only:
        - master
  5. Push your project to GitLab, which triggers a CI job pipeline. This pipeline runs the gitlab-terraform init, gitlab-terraform validate, and gitlab-terraform plan commands.

The output from the above terraform commands should be viewable in the job logs.

WARNING: Like any other job artifact, Terraform plan data is viewable by anyone with the Guest role on the repository. Neither Terraform nor GitLab encrypts the plan file by default. If your Terraform plan includes sensitive data such as passwords, access tokens, or certificates, GitLab strongly recommends encrypting plan output or modifying the project visibility settings.

Example project

See this reference project using GitLab and Terraform to deploy a basic AWS EC2 in a custom VPC.

Using a GitLab-managed Terraform state backend as a remote data source

You can use a GitLab-managed Terraform state as a Terraform data source. To use your existing Terraform state backend as a data source, provide the following details as Terraform input variables:

  • address: The URL of the remote state backend you want to use as a data source. For example, https://gitlab.com/api/v4/projects/<TARGET-PROJECT-ID>/terraform/state/<TARGET-STATE-NAME>.
  • username: The username to authenticate with the data source. If you are using a Personal Access Token for authentication, this is your GitLab username. If you are using GitLab CI, this is 'gitlab-ci-token'.
  • password: The password to authenticate with the data source. If you are using a Personal Access Token for authentication, this is the token value. If you are using GitLab CI, it is the contents of the ${CI_JOB_TOKEN} CI/CD variable.

An example setup is shown below:

  1. Create a file named example.auto.tfvars with the following contents:

    example_remote_state_address=https://gitlab.com/api/v4/projects/<TARGET-PROJECT-ID>/terraform/state/<TARGET-STATE-NAME>
    example_username=<GitLab username>
    example_access_token=<GitLab Personal Access Token>
  2. Define the data source by adding the following code block in a .tf file (such as data.tf):

    data "terraform_remote_state" "example" {
      backend = "http"
    
      config = {
        address = var.example_remote_state_address
        username = var.example_username
        password = var.example_access_token
      }
    }

Outputs from the data source can now be referenced in your Terraform resources using data.terraform_remote_state.example.outputs.<OUTPUT-NAME>.

You need at least the Developer role in the target project to read the Terraform state.

Migrating to GitLab-managed Terraform state

Terraform supports copying the state when the backend is changed or reconfigured. This can be useful if you need to migrate from another backend to GitLab-managed Terraform state. Using a local terminal is recommended to run the commands needed for migrating to GitLab-managed Terraform state.

The following example demonstrates how to change the state name, the same workflow is needed to migrate to GitLab-managed Terraform state from a different state storage backend.

Setting up the initial backend

PROJECT_ID="<gitlab-project-id>"
TF_USERNAME="<gitlab-username>"
TF_PASSWORD="<gitlab-personal-access-token>"
TF_ADDRESS="https://gitlab.com/api/v4/projects/${PROJECT_ID}/terraform/state/old-state-name"

terraform init \
  -backend-config=address=${TF_ADDRESS} \
  -backend-config=lock_address=${TF_ADDRESS}/lock \
  -backend-config=unlock_address=${TF_ADDRESS}/lock \
  -backend-config=username=${TF_USERNAME} \
  -backend-config=password=${TF_PASSWORD} \
  -backend-config=lock_method=POST \
  -backend-config=unlock_method=DELETE \
  -backend-config=retry_wait_min=5
Initializing the backend...

Successfully configured the backend "http"! Terraform will automatically
use this backend unless the backend configuration changes.

Initializing provider plugins...

Terraform has been successfully initialized!

You may now begin working with Terraform. Try running "terraform plan" to see
any changes that are required for your infrastructure. All Terraform commands
should now work.

If you ever set or change modules or backend configuration for Terraform,
re-run this command to reinitialize your working directory. If you forget, other
commands will detect it and remind you to do so if necessary.

Changing the backend

Now that terraform init has created a .terraform/ directory that knows where the old state is, you can tell it about the new location:

TF_ADDRESS="https://gitlab.com/api/v4/projects/${PROJECT_ID}/terraform/state/new-state-name"

terraform init \
  -migrate-state \
  -backend-config=address=${TF_ADDRESS} \
  -backend-config=lock_address=${TF_ADDRESS}/lock \
  -backend-config=unlock_address=${TF_ADDRESS}/lock \
  -backend-config=username=${TF_USERNAME} \
  -backend-config=password=${TF_PASSWORD} \
  -backend-config=lock_method=POST \
  -backend-config=unlock_method=DELETE \
  -backend-config=retry_wait_min=5
Initializing the backend...
Backend configuration changed!

Terraform has detected that the configuration specified for the backend
has changed. Terraform will now check for existing state in the backends.


Acquiring state lock. This may take a few moments...
Do you want to copy existing state to the new backend?
  Pre-existing state was found while migrating the previous "http" backend to the
  newly configured "http" backend. No existing state was found in the newly
  configured "http" backend. Do you want to copy this state to the new "http"
  backend? Enter "yes" to copy and "no" to start with an empty state.

  Enter a value: yes


Successfully configured the backend "http"! Terraform will automatically
use this backend unless the backend configuration changes.

Initializing provider plugins...

Terraform has been successfully initialized!

You may now begin working with Terraform. Try running "terraform plan" to see
any changes that are required for your infrastructure. All Terraform commands
should now work.

If you ever set or change modules or backend configuration for Terraform,
rerun this command to reinitialize your working directory. If you forget, other
commands will detect it and remind you to do so if necessary.

If you type yes, it copies your state from the old location to the new location. You can then go back to running it in GitLab CI/CD.

Managing state files

Introduced in GitLab 13.8.

Users with at least the Developer role can view the state files attached to a project at Infrastructure > Terraform. Users with the Maintainer role can perform commands on the state files. The user interface contains these fields:

Terraform state list

  • Name: The name of the environment, with a locked ({lock}) icon if the state file is locked.
  • Pipeline: A link to the most recent pipeline and its status.
  • Details: Information about when the state file was created or changed.
  • Actions: Actions you can take on the state file, including copying the terraform init command, downloading, locking, unlocking, or removing the state file and versions.

NOTE: Additional improvements to the graphical interface for managing state files are planned.

Manage individual Terraform state versions

Introduced in GitLab 13.4.

Individual state versions can be managed using the GitLab REST API.

Users with the Developer role can retrieve state versions using their serial number. To retrieve a version:

curl --header "Private-Token: <your_access_token>" "https://gitlab.example.com/api/v4/projects/<your_project_id>/terraform/state/<your_state_name>/versions/<version-serial>"

Users with the Maintainer role can remove state versions using their serial number. To remove a version:

curl --header "Private-Token: <your_access_token>" --request DELETE "https://gitlab.example.com/api/v4/projects/<your_project_id>/terraform/state/<your_state_name>/versions/<version-serial>"

Remove a state file

Users with at least the Maintainer role can use the following options to remove a state file:

  • GitLab UI: Go to Infrastructure > Terraform. In the Actions column, click the vertical ellipsis ({ellipsis_v}) button and select Remove state file and versions.

  • GitLab REST API: You can remove a state file by making a request to the REST API. For example:

    curl --header "Private-Token: <your_access_token>" --request DELETE "https://gitlab.example.com/api/v4/projects/<your_project_id>/terraform/state/<your_state_name>"
  • GitLab GraphQL API.

Remove a state file with the GitLab GraphQL API

You can remove a state file by making a GraphQL API request. For example:

mutation deleteState {
  terraformStateDelete(input: { id: "<global_id_for_the_state>" }) {
    errors
  }
}

You can obtain the <global_id_for_the_state> by querying the list of states:

query ProjectTerraformStates {
  project(fullPath: "<your_project_path>") {
    terraformStates {
      nodes {
        id
        name
      }
    }
  }
}

For those new to the GitLab GraphQL API, read Getting started with GitLab GraphQL API.

Troubleshooting

Unable to lock Terraform state files in CI jobs for terraform apply using a plan created in a previous job

When passing -backend-config= to terraform init, Terraform persists these values inside the plan cache file. This includes the password value.

As a result, to create a plan and later use the same plan in another CI job, you might get the error Error: Error acquiring the state lock errors when using -backend-config=password=$CI_JOB_TOKEN. This happens because the value of $CI_JOB_TOKEN is only valid for the duration of the current job.

As a workaround, use http backend configuration variables in your CI job, which is what happens behind the scenes when following the Get started using GitLab CI instructions.

Error: "address": required field is not set

By default, we set TF_ADDRESS to ${CI_API_V4_URL}/projects/${CI_PROJECT_ID}/terraform/state/${TF_STATE_NAME}. If you don't set TF_STATE_NAME or TF_ADDRESS in your job, the job fails with the error message Error: "address": required field is not set.

To resolve this, ensure that either TF_ADDRESS or TF_STATE_NAME is accessible in the job that returned the error:

  1. Configure the CI/CD environment scope for the job.
  2. Set the job's environment, matching the environment scope from the previous step.